Paginate Results

To page through a large query result, you can get your desired part in a "paginated" manner.

Paginate query results

In TiDB, you can paginate query results using the LIMIT statement. For example:

SELECT * FROM table_a t ORDER BY gmt_modified DESC LIMIT offset, row_count;

offset indicates the beginning number of records and row_count indicates the number of records per page. TiDB also supports LIMIT row_count OFFSET offset syntax.

When pagination is used, it is recommended that you sort query results with the ORDER BY statement unless there is a need to display data randomly.

• SQL
• Java

SELECT *
FROM books
ORDER BY published_at DESC
LIMIT 0, 10;

public List<Book> getLatestBooksPage(Long pageNumber, Long pageSize) throws SQLException {
    pageNumber = pageNumber < 1L ? 1L : pageNumber;
    pageSize = pageSize < 10L ? 10L : pageSize;
    Long offset = (pageNumber - 1) * pageSize;
    Long limit = pageSize;
    List<Book> books = new ArrayList<>();
    try (Connection conn = ds.getConnection()) {
        PreparedStatement stmt = conn.prepareStatement("""
        SELECT id, title, published_at
        FROM books
        ORDER BY published_at DESC
        LIMIT ?, ?;
        """);
        stmt.setLong(1, offset);
        stmt.setLong(2, limit);
        ResultSet rs = stmt.executeQuery();
        while (rs.next()) {
            Book book = new Book();
            book.setId(rs.getLong("id"));
            book.setTitle(rs.getString("title"));
            book.setPublishedAt(rs.getDate("published_at"));
            books.add(book);
        }
    }
    return books;
}

Paging batches for single-field primary key tables

Usually, you can write a pagination SQL statement using a primary key or unique index to sort results and the offset keyword in the LIMIT clause to split pages by a specified row count. Then the pages are wrapped into independent transactions to achieve flexible paging updates. However, the disadvantage is also obvious. As the primary key or unique index needs to be sorted, a larger offset consumes more computing resources, especially in the case of a large volume of data.

The following introduces a more efficient paging batching method:

• SQL
• Java

SELECT
    floor((t.row_num - 1) / 1000) + 1 AS page_num,
    min(t.id) AS start_key,
    max(t.id) AS end_key,
    count(*) AS page_size
FROM (
    SELECT id, row_number() OVER (ORDER BY id) AS row_num
    FROM books
) t
GROUP BY page_num
ORDER BY page_num;

+----------+------------+------------+-----------+
| page_num | start_key  | end_key    | page_size |
+----------+------------+------------+-----------+
|        1 |     268996 |  213168525 |      1000 |
|        2 |  213210359 |  430012226 |      1000 |
|        3 |  430137681 |  647846033 |      1000 |
|        4 |  647998334 |  848878952 |      1000 |
|        5 |  848899254 | 1040978080 |      1000 |
...
|       20 | 4077418867 | 4294004213 |      1000 |
+----------+------------+------------+-----------+
20 rows in set (0.01 sec)

DELETE FROM books
WHERE
    id BETWEEN 268996 AND 213168525
ORDER BY id;

public class PageMeta<K> {
    private Long pageNum;
    private K startKey;
    private K endKey;
    private Long pageSize;

    // Skip the getters and setters.

}

public class BookDAO {
    public List<PageMeta<Long>> getPageMetaList() throws SQLException {
        List<PageMeta<Long>> pageMetaList = new ArrayList<>();
        try (Connection conn = ds.getConnection()) {
            Statement stmt = conn.createStatement();
            ResultSet rs = stmt.executeQuery("""
            SELECT
                floor((t.row_num - 1) / 1000) + 1 AS page_num,
                min(t.id) AS start_key,
                max(t.id) AS end_key,
                count(*) AS page_size
            FROM (
                SELECT id, row_number() OVER (ORDER BY id) AS row_num
                FROM books
            ) t
            GROUP BY page_num
            ORDER BY page_num;
            """);
            while (rs.next()) {
                PageMeta<Long> pageMeta = new PageMeta<>();
                pageMeta.setPageNum(rs.getLong("page_num"));
                pageMeta.setStartKey(rs.getLong("start_key"));
                pageMeta.setEndKey(rs.getLong("end_key"));
                pageMeta.setPageSize(rs.getLong("page_size"));
                pageMetaList.add(pageMeta);
            }
        }
        return pageMetaList;
    }

    public void deleteBooksByPageMeta(PageMeta<Long> pageMeta) throws SQLException {
        try (Connection conn = ds.getConnection()) {
            PreparedStatement stmt = conn.prepareStatement("DELETE FROM books WHERE id >= ? AND id <= ?");
            stmt.setLong(1, pageMeta.getStartKey());
            stmt.setLong(2, pageMeta.getEndKey());
            stmt.executeUpdate();
        }
    }
}

List<PageMeta<Long>> pageMetaList = bookDAO.getPageMetaList();
if (pageMetaList.size() > 0) {
    bookDAO.deleteBooksByPageMeta(pageMetaList.get(0));
}

List<PageMeta<Long>> pageMetaList = bookDAO.getPageMetaList();
pageMetaList.forEach((pageMeta) -> {
    try {
        bookDAO.deleteBooksByPageMeta(pageMeta);
    } catch (SQLException e) {
        e.printStackTrace();
    }
});

This method significantly improves the efficiency of batch processing by avoiding wasting computing resources caused by frequent data sorting operations.

Paging batches for composite primary key tables

Non-clustered index table

For non-clustered index tables (also known as "non-index-organized tables"), the internal field _tidb_rowid can be used as a pagination key, and the pagination method is the same as that of single-field primary key tables.

For example:

SELECT
    floor((t.row_num - 1) / 1000) + 1 AS page_num,
    min(t._tidb_rowid) AS start_key,
    max(t._tidb_rowid) AS end_key,
    count(*) AS page_size
FROM (
    SELECT _tidb_rowid, row_number () OVER (ORDER BY _tidb_rowid) AS row_num
    FROM users
) t
GROUP BY page_num
ORDER BY page_num;

The result is as follows:

+----------+-----------+---------+-----------+
| page_num | start_key | end_key | page_size |
+----------+-----------+---------+-----------+
|        1 |         1 |    1000 |      1000 |
|        2 |      1001 |    2000 |      1000 |
|        3 |      2001 |    3000 |      1000 |
|        4 |      3001 |    4000 |      1000 |
|        5 |      4001 |    5000 |      1000 |
|        6 |      5001 |    6000 |      1000 |
|        7 |      6001 |    7000 |      1000 |
|        8 |      7001 |    8000 |      1000 |
|        9 |      8001 |    9000 |      1000 |
|       10 |      9001 |    9990 |       990 |
+----------+-----------+---------+-----------+
10 rows in set (0.00 sec)

Clustered index table

For clustered index tables (also known as "index-organized tables"), you can use the concat function to concatenate values of multiple columns as a key, and then use a window function to query the paging information.

It should be noted that the key is a string at this time, and you must ensure that the length of the string is always the same, to obtain the correct start_key and end_key in the slice through the min and max aggregation function. If the length of the field for string concatenation is not fixed, you can use the LPAD function to pad it.

For example, you can implement a paging batch for the data in the ratings table as follows:

Create the meta information table by using the following statement. As the key concatenated by book_id and user_id, which are bigint types, is unable to convert to the same length, the LPAD function is used to pad the length with 0 according to the maximum bits 19 of bigint.

SELECT
    floor((t1.row_num - 1) / 10000) + 1 AS page_num,
    min(mvalue) AS start_key,
    max(mvalue) AS end_key,
    count(*) AS page_size
FROM (
    SELECT
        concat('(', LPAD(book_id, 19, 0), ',', LPAD(user_id, 19, 0), ')') AS mvalue,
        row_number() OVER (ORDER BY book_id, user_id) AS row_num
    FROM ratings
) t1
GROUP BY page_num
ORDER BY page_num;

The result is as follows:

+----------+-------------------------------------------+-------------------------------------------+-----------+
| page_num | start_key                                 | end_key                                   | page_size |
+----------+-------------------------------------------+-------------------------------------------+-----------+
|        1 | (0000000000000268996,0000000000092104804) | (0000000000140982742,0000000000374645100) |     10000 |
|        2 | (0000000000140982742,0000000000456757551) | (0000000000287195082,0000000004053200550) |     10000 |
|        3 | (0000000000287196791,0000000000191962769) | (0000000000434010216,0000000000237646714) |     10000 |
|        4 | (0000000000434010216,0000000000375066168) | (0000000000578893327,0000000002167504460) |     10000 |
|        5 | (0000000000578893327,0000000002457322286) | (0000000000718287668,0000000001502744628) |     10000 |
...
|       29 | (0000000004002523918,0000000000902930986) | (0000000004147203315,0000000004090920746) |     10000 |
|       30 | (0000000004147421329,0000000000319181561) | (0000000004294004213,0000000003586311166) |      9972 |
+----------+-------------------------------------------+-------------------------------------------+-----------+
30 rows in set (0.28 sec)

To delete all rating records on page 1, replace the start_key and end_key with values of page 1 in the above result:

SELECT * FROM ratings
WHERE
    (book_id > 268996 AND book_id < 140982742)
    OR (
        book_id = 268996 AND user_id >= 92104804
    )
    OR (
        book_id = 140982742 AND user_id <= 374645100
    )
ORDER BY book_id, user_id;